Open Floor Protocol   ›   Introduction

Introduction

Get started with the Open Floor Protocol (OFP)

OFP is an open protocol that standardizes how conversational agents communicate and collaborate with each other. OFP has been defined and is under development by the Linux Foundation AI & Data Foundation Open Voice Interoperability Initiative (https://voiceinteroperability.ai).

Think of OFP like a conference room protocol for AI agents. Just as people in a meeting follow certain conventions to speak, listen, and collaborate effectively, OFP provides a standardized way for AI agents (and humans) to gather around a shared conversational "floor" and work together to solve problems.

The Open Floor Protocol enables any conversational agent that can generate and send message envelopes to interoperate with any other Open Floor-compliant agent, regardless of the technology or architecture on which that other agent is based.

openfloor.dev is a developer-focused resource hub designed to make implementing the Open Floor Protocol as straightforward as possible for developers of all skill levels. The site provides comprehensive documentation, interactive examples, testing tools, and ready-to-use code samples to help you build, validate, and deploy Open Floor-compliant conversational agents.

Key Characteristics

  • Multi-party conversation model: Uses the metaphor of participants gathering around a shared floor, similar to a conference room or meeting space
  • Vendor-independent: Open standard that enables interoperability between agents built on different technology stacks
  • Mixed participation: Supports conversations with any combination of human and autonomous agents as participants.

Core Components

  • Conversation Envelopes: Universal JSON message format for agent communication
  • Dialog Events: Standardized structure for representing linguistic events and utterances
  • Assistant Manifests: "Curriculum vitae" describing agent capabilities and identity

Communication Patterns

  • Delegation: Transferring control from one dialog agent to another.
  • Channeling: Acting as an intermediary, passing messages between agents without altering the message content.
  • Mediation: Interacting with multiple agents behind the scenes to achieve a goal before responding to the user.
  • Orchestration: A convener agent manages communication around an open floor for agents and human conversants to collaborate to achieve a goal.

Example Use Cases

  1. Basic Communication: A user interacts with multiple autonomous assistants sequentially, managed by a floor manager converting and forwarding messages.
  2. Delegated Tasks: A dialog agent delegates part of the conversation or task to another agent, retaining overall control and context.
  3. Discovery and Recommendations: Agents can query other agents to determine their capabilities and seek recommendations. Agents can propose themselves or others for specific tasks.
  4. Multi-agent Interaction: Multiple agents and/or users gather round a floor under a convener agent and collaborate to achieve a task.

Basic Flow

The following video shows a basic conversation using the Open Floor Protocol.